TCSinger 2 generates zero-shot singing voices in nine languages with style transfer from audio prompts and multi-level style control from natural language prompts, using blurred boundary encoders, contrastive prompt alignment, and a flow-based transformer.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
TCSinger 2: Customizable Multilingual Zero-shot Singing Voice Synthesis
TCSinger 2 generates zero-shot singing voices in nine languages with style transfer from audio prompts and multi-level style control from natural language prompts, using blurred boundary encoders, contrastive prompt alignment, and a flow-based transformer.